China's AI Race: Resolve Over Money, but Chip Gap Remains

Bruce Liu, CEO of Esoterica Capital, said in the newsletter that if China had one dollar left, "that dollar is going to be spent on AI rather than real estate." He said China's goal is to be self-sufficient in AI and not rely on the U.S., even if it does not have the best AI in the world. National policy and district-level subsidies reflect that ambition, he added.

The U.S. capital advantage is substantial. According to Alexander Kheder, TMT analyst at BMI, a Fitch Solutions unit, private-sector AI investment in the U.S. is about 23 times greater than in mainland China. Nvidia has gathered Wall Street backers for $500 billion in AI development financing, demonstrating that edge. Unless Beijing lets Chinese AI firms tap external non-state capital, Kheder said, "this financing asymmetry will remain one of the most durable structural explanations for US leadership."

Meanwhile, Chinese companies have released AI models with similar capabilities at lower prices, drawing global interest even after DeepSeek raised prices over the weekend. But running those models still requires chips, an area where China lags. Clifford Kurz, director at S&P Global Ratings, said Beijing "could announce even more financial support," but "if they don't have the chips, what's the point of support? There's nothing to finance."

Kurz noted that Huawei offers roughly one-eighth of Nvidia's computing capacity, mostly outside China. Each of Huawei's most advanced Ascend 950 chips has about 13% of the computing power of one Nvidia GB300 chip, he said. Nvidia's even more powerful Vera Rubin chip is due this year, while Huawei has compensated by stacking more chips together. But Kurz expects Huawei to produce just 1.35 million advanced AI chips this year, far fewer than the most conservative estimate of 6 million Nvidia chips.

The gap could narrow quickly, CNBC said. Huawei and other Chinese AI supply-chain companies have closed in over just a few years, and China has low electricity costs and is courting AI talent. In June, Beijing released a three-year plan for computing-power infrastructure, and last month said the buildout could attract 4 trillion yuan in capital through 2030.

Investors are taking note. William Chow, deputy group CEO of Raffles Family Office, said China's domestic semiconductor push creates a "parallel" opportunity rather than competition for capital flowing to U.S. tech. He said clients this year are more focused on entry prices for AI investments, and that Nvidia's financing plan shifts risk from equity to credit, making diversification more important.

Financing methods also differ. Zhu He, senior fellow at the CF40 Institute, told CNBC (in Mandarin translated by CNBC) that leading Chinese companies have no significant plans for large-scale debt issuance for AI. Most use equity financing and internal funds, he said, with telecommunications giants and internet companies investing.

Regardless of country, the scale of money needed for AI marks a departure from the asset-light models that drove business over the past two decades, said Esoterica's Liu. "Hyperscalers need to spend to get ahead." AI still faces a commercialization test, he added. U.S. companies have invested heavily to create the smartest models, while China's focus is integrating AI across industries. "The AI rivalry is about applications based on the full AI stack," said Winston Ma, adjunct professor of law at New York University. Whoever finds the right formula is poised to win.